|Publication number||US6915316 B1|
|Application number||US 10/736,215|
|Publication date||Jul 5, 2005|
|Filing date||Dec 15, 2003|
|Priority date||Apr 3, 2001|
|Also published as||US6668264|
|Publication number||10736215, 736215, US 6915316 B1, US 6915316B1, US-B1-6915316, US6915316 B1, US6915316B1|
|Inventors||Hugo Patterson, Michael Federwisch|
|Original Assignee||Network Appliance, Inc.|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (16), Non-Patent Citations (3), Referenced by (25), Classifications (23), Legal Events (2)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This application hereby incorporates by reference and is a continuation of application Ser. No. 09/825,855, filed Apr. 3, 2001, now U.S. Pat. No. 6,668,264.
1. Field of the Invention
This invention relates to data storage systems; in particular, the invention relates to synchronization of source and target volumes in a mirrored storage system.
2. Related Art
Snapshots and multiple volumes are frequently used to prevent data loss when a data storage drive fails in a file system. Such snapshots “capture” the contents of the files and directories in a volume at a particular point in time in order to recover earlier versions of a file following an unintended deletion or modification. Such snapshots can also be copied to one or more volumes, which then can be used as a mirror or a collection or mirrors and which can provide a back-up copy of the file system. When used in this way, the mirror can be referred to as a target volume. In general, a target volume is a “read-only” volume that contains a set of data that is equivalent to the set of data on an original source volume. Such target volumes can be written to only by the original source volume.
A target volume may be updated periodically with respect to a source volume by looking to the most recent snapshot that the target and source have in common and using that snapshot as a consistency point (CP). The file blocks in the most recent common snapshot and the file blocks of a new snapshot are compared. The set of differences resulting from this comparison are written to the less up-to-date volume. In this way, both source and target volumes maintain equivalent sets of file blocks.
A source volume may become unavailable due to a failure of the source volume or to a failed connection to the source volume. Under such conditions, it is advantageous to temporarily use the target volume as a source volume by designating it as a “read/write” volume. User I/Os are directed to write to the target volume while the original source volume is unavailable.
One problem with writing to a target volume is that it may cause the target volume to contain data not found in the original source volume. A partial solution to this problem involves transferring data from the target to the source once the source is restored. However this is undesirable because it requires diversion of computational resources and filer overhead.
Accordingly, it would be advantageous to provide an improved technique for quickly and efficiently updating source and target volumes after a target volume has been written to. This is achieved in an embodiment of the invention that addresses the foregoing deficiencies.
The invention provides an improved method and apparatus for quickly and efficiently updating an original source volume and an original target volume after the original target volume has been used as a source volume in a file system. One or more snapshots are used to compare data included in the source and target volume. Instead of transferring the entire volume, only the data that is missing from a source and a target volume is transferred.
In a first aspect of the invention, a target volume becomes synchronized with a source volume after the target has been written to by an entity other than the source. Synchronization is a two phase process. In the first phase, a target volume provides the source volume with a list of snapshots and associated snapshot numbers used to determine the sequence of the snapshots on the list. The source volume compares its own list of snapshot numbers with the list of the target volume's snapshot numbers and determines the newest common snapshot. This newest common snapshot is a consistency point between the target volume and the source volume. The source volume sends the target volume a set of snapshot numbers describing the newest common snapshot and the target volume reverts back to this snapshot. In the second phase, file system software identifies all the data blocks contained in any one or more of the snapshots of the source volume. This file system software also identifies all the data blocks in any one or more of the snapshots of the target volume using the data sent by the target volume to the source volume as described supra. A set of data blocks that are included in the source volume and not included in the target volume is generated. This can be accomplished by making a comparison based on logical differences, generating a virtual or actual list or other techniques known in the art.
File system software synchronizes the target volume with the source volume. First, the file system software removes snapshots from a target volume if the snapshots are not included in the source volume's snapshot list. Second, the file system software adds the set of data blocks identified above (that is the set of data blocks that are included in the source volume and not included in the target volume) to its memory. Lastly, the file system software adds snapshots to the target volume if the snapshots are included in the source volume's snapshot list and not in the target volume's snapshot list. At this point, the target volume includes the data blocks that are present on the source volume.
In a second aspect of the invention, the roles of the target volume and source volume are reversed and the process described supra is performed again so as to synchronize source volume with the target volume. This is necessary because the target volume may include data blocks not included in the source volume. After both source and target volumes are synchronized, the target volume stops being written to and the source once again is used as the active file. This is accomplished by 1) designating the target as a read-only volume, 2) designating the source as a read/write volume, and 3) redirecting users' I/O's back to the source volume.
In a preferred embodiment, sources and volumes can be synchronized dynamically, using a WAFL (Write Anywhere File Layout) system using RAID (Redundant Arrays of Independent Disks) architecture. However, various other types of file systems involving redundant copies of data can also be used.
The following terms are related to aspects of the invention as described below. The general meanings of these terms are exemplary and in no way limiting.
Source volume—in general, the term “source volume” refers to a read/write volume.
Target volume—in general, the term “target volume” refers to a read-only volume that is used to back-up other data. However, in the event that a source volume becomes unavailable, a target volume may be designated as “read/write” and used as a source.
Synchronize—in general, the term “synchronize” refers to the process of conforming a first set of snapshots to a second set of snapshots.
In the following description, a preferred embodiment of the invention is described with regard to preferred process steps and data structures. Embodiments of the invention can be implemented using general-purpose processors or special purpose processors operating under program control, or other circuits adapted to particular process steps and data structures described herein. Implementation of the process steps and structures described herein would not require undue experimentation or further invention.
A system for synchronizing a target volume to a source volume (shown by general character reference 100) includes a file system 110, upon which resides one or more source volumes 120, one or more target volumes 140, and logic 115. In a preferred embodiment, the file system 110 is part of a larger computer system including a memory and a processor. The file system 110 is coupled to an I/O port 105.
The source volume 120 includes a set of data blocks 125 and a set of snapshots 130. The set of data blocks 125 both data and meta-data.
The set of snapshots 130 includes individual snapshots that correspond to the set of data blocks 125 at various points in time. A snapshot includes a map of blocks at a consistent point in the file system, but preferably not the blocks themselves. The individual snapshots include snapshot numbers which refer to the relative age of the snapshot. In a preferred embodiment, the higher snapshot numbers correspond to more recent snapshots and lower snapshot numbers correspond to older snapshots. Although the snapshot numbers shown in
In a preferred embodiment, the source volume 120 is a read/write volume that receives user I/Os 105. A system 100 may include a plurality of source volumes 120. In the event that a source volume 120 becomes unavailable, a target volume 140 may be temporarily used as a source volume 120.
The target volume 140 includes a set of data blocks 145 and a set of snapshots 150. Similar to the set of data blocks 125 included in the source volume 120, the set of data blocks 145 includes individual data blocks, indirect data blocks, and double indirect data blocks. The set of snapshots 130 (which are themselves data blocks) includes individual snapshots of the set of data blocks 145 at various points in time. The individual snapshots include snapshot numbers relating to the relative age of a snapshot Generally, the highest snapshot number corresponds to the most recent snapshot. Similar to the snapshot numbers included in the target volume 120, these snapshot numbers need not be a uniform sequence. Both the number and type of data blocks in the set of data blocks 145, as well as the number and sequence of snapshots in the set of snapshots 150, are exemplary and in no way limiting.
In a preferred embodiment, the target volume 140 is a read-only volume that is preferably used to replicate data from the source volume 120. When used as such, user I/Os 105 are not directed to the target volume 140, but rather to the source volume 120. A system 100 may include a plurality number of source volumes 120 and target volumes 140, such that the source volumes 120 mirror data to the redundant target volumes 140.
The logic 115 provides a technique for synchronizing a source volume 120 to a target volume 140, a target volume 140 to a source volume 120, or both. Generally, this technique is used when a source volume 120 is taken off line and a target volume 140 is temporarily used in it's place or when the target volume 140 is written to by any entity s other than the source volume 120.
Method of Use
A method for synchronizing a target volume to a source volume (shown by general character reference 200) is performed by a system 100. Although a method 200 is described serially, steps of a method 200 can be performed by separate elements in conjunction or in parallel, whether asynchronously, in a pipelined manner, or otherwise. There is no particular requirement that a method 200 be performed in the same order in which this description lists the steps, except where so indicated.
At a flow point 205, a system 100 is ready to perform a method 200 and synchronize a target volume 140 to a source volume 120. The method 200 is preferably performed after a target volume 140 has been made writable.
In a step 210, the logic 115 identifies a set of snapshots 150 associated with a target volume 140 and copies the identifiers associated with those snapshots 150 to a source volume 120. For example, if the set of snapshots 150 includes snapshot numbers 1, 3, 4 and 6, those particular snapshots numbers are copied to the source volume 120.
At a step 220, the logic 115 compares the identifiers associated with the set of snapshots 150 identified in step 210 with the set of snapshots 130. The most recent snapshot that is common to both sets is selected. For example, if the identifiers include snapshot numbers 1, 3, 4 and 6 and the set of snapshots 130 includes snapshot numbers 1, 2, 4, and 5, then the most recent snapshot common to both sets is snapshot number 4.
At a step 225, the identifier associated with the most recent common snapshot (that is, the snapshot selected in step 220) is copied from the source volume 120 to the target volume 140. During this step, an alert may be sounded, informing the user I/O 105 that some of the data blocks unique to the snapshots newer than the newest common snapshot may be lost and a prompt for “confirmation” or “abort” may be issued. If there is a confirmation, the target volume 140 reverts back to the most recent snapshot selected in step 220. This reversion may be referred to as a “SnapRestore”.
At a step 228, the source volume 120 generates a new snapshot of itself. This snapshot is used to preserve the set of data blocks 125 at the source volume 120 and to determine the incremental transfer of data blocks between the source volume 120 and target volume 140. Simultaneously, the target volume 140 is designated as a read only volume.
In a preferred embodiment, the source volume 120 may continue receiving reads and writes from clients.
At a step 229, the snapshots that are not included in the source volume 120 are removed from the set of snapshot 150 on the target volume 140. In this way, the target volume becomes synchronized with respect to snapshots that are not present on the source volume 120.
At a step 230, the union of the data blocks in the set of snapshots 130 is computed. This union will preferably include available and allocated data blocks from any one or more target snapshots 130. This step is preferably performed by the source volume 120. Steps 235 and 245 occur simultaneously with step 230.
At a step 235, the union of the blocks in the set of snapshots 150 is computed. This union will preferably include available and allocated data blocks from any one or more target snapshots 150. This step is preferably performed by the source volume 120. Step 235 is performed at the same time as steps 230 and 245.
At a step 245, difference between the unions calculated in steps 230 and 235 is calculated. This difference represents the blocks to be transferred. These blocks are copied from the source volume 120 to the target volume 140. In this way, the target volume 140 becomes synchronized with respect to blocks present on the source volume 120.
At a flow point 260, the target volume 140 is synchronized with respect to the source volume 120 and the method 200 is complete.
A method for aligning a target volume and a source volume to each other (shown by general character reference 300) is performed by a system 100. Although the method 300 is described serially, the steps of method 300 can be performed by separate elements in conjunction or in parallel, whether asynchronously, in a pipelined manner, or otherwise. There is no particular requirement that method 300 be performed in the same order, in which this description lists the steps, except where so indicated.
At a flow point 310, system 100 is ready to perform a method 300. The method 300 compensates for the asynchronous nature of mirroring data. At this time, both the source volume and target volume are taken off line. This prevents users form writing to either volume during the transition.
At a step 315, the target volume 120 becomes re-synchronized with the source volume 140 as shown in FIG. 2. Upon completion of step 315, the target volume 120 includes information stored on the source volume 140. However, given the asynchronous nature of mirroring from the source volume 140 to the target volume 120, the source volume 140 may include a little bit of data that is not present on the target volume 120. This inconsistency is rectified in the following steps.
At a step 320, the target volume 120 is designated as read/write. However, user I/Os are not directed to it at this time.
At a step 330, the method 200 is performed again such that the roles of the target volume and source volume are reversed so that data on the source volume 140 is mirrored to the target volume 120. Upon completion of this step, the target volume includes all of the date that was originally on the source volume 140 and does not include any data that was not on the source volume 140.
At a step 335, the target volume 120 is designated as a read only and the source volume 140 is designated read/write. User I/Os 105 are directed to the source volume 140.
At a flow point 340, the method 300 is complete.
Although preferred embodiments are disclosed herein, many variations are possible which remain within the concept, scope, and spirit of the invention, and these variations would become clear to those skilled in the art after perusal of this application.
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US5835953||Nov 8, 1996||Nov 10, 1998||Vinca Corporation||Backup system that takes a snapshot of the locations in a mass storage device that has been identified for updating prior to updating|
|US5907672 *||Oct 4, 1995||May 25, 1999||Stac, Inc.||System for backing up computer disk volumes with error remapping of flawed memory addresses|
|US5996086 *||Oct 14, 1997||Nov 30, 1999||Lsi Logic Corporation||Context-based failover architecture for redundant servers|
|US6000039 *||Jul 17, 1997||Dec 7, 1999||Hitachi, Ltd.||Data reconstruction method and system wherein timing of data reconstruction is controlled in accordance with conditions when a failure occurs|
|US6078932 *||Jan 13, 1998||Jun 20, 2000||International Business Machines Corporation||Point-in-time backup utilizing multiple copy technologies|
|US6119244 *||Aug 25, 1998||Sep 12, 2000||Network Appliance, Inc.||Coordinating persistent status information with multiple file servers|
|US6131088 *||May 18, 1998||Oct 10, 2000||Charles E. Hill & Associates, Inc.||Electronic catalog system and method|
|US6212531 *||Jan 13, 1998||Apr 3, 2001||International Business Machines Corporation||Method for implementing point-in-time copy using a snapshot function|
|US6360330||Mar 31, 1998||Mar 19, 2002||Emc Corporation||System and method for backing up data stored in multiple mirrors on a mass storage subsystem under control of a backup server|
|US6377951 *||Aug 6, 1999||Apr 23, 2002||Punch Networks Corporation||On-line database updating network system and method|
|US6453325||Feb 3, 1997||Sep 17, 2002||International Business Machines Corporation||Method and means for backup and restoration of a database system linked to a system for filing data|
|US6529921||Jun 29, 1999||Mar 4, 2003||Microsoft Corporation||Dynamic synchronization of tables|
|US6529944 *||Jan 6, 1999||Mar 4, 2003||Emc Corporation||Host system for remote control of mass storage volumes using cascading commands|
|US6643671 *||Aug 27, 2001||Nov 4, 2003||Storage Technology Corporation||System and method for synchronizing a data copy using an accumulation remote copy trio consistency group|
|US6668264 *||Apr 3, 2001||Dec 23, 2003||Network Appliance, Inc.||Resynchronization of a target volume with a source volume|
|WO1994029796A1||Jun 2, 1994||Dec 22, 1994||Network Appliance Corp||A method for allocating files in a file system integrated with a raid disk sub-system|
|1||*||Beltrane et al., "An APL interface to the open system interconnection for labor scale application systems", ACM, 1984, pp. 25-31.|
|2||*||Leslie Jill Miller, The ISO reference model of open system interconnection: A first tutorial, ACM'81, Nov. 9-11.|
|3||*||M. F. Dolan "Aminimal duplex connection capability in the top three layers of the OSI reference model", ACM, 1984, pp. 195-200.|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7240080 *||Jul 30, 2003||Jul 3, 2007||International Business Machines Corporation||Method and apparatus for determining using least recently used protocol if one or more computer files should be written to one or more information storage media and synchronously providing one or more computer files between first and storage devices|
|US7644109||Jan 9, 2006||Jan 5, 2010||Netapp, Inc.||System and method for storage of snapshot metadata in a remote file|
|US7644112||Feb 1, 2007||Jan 5, 2010||International Business Machines Corporation||Apparatus, system, and method for initializing a synchronized remote database|
|US7685178||Oct 31, 2006||Mar 23, 2010||Netapp, Inc.||System and method for examining client generated content stored on a data container exported by a storage system|
|US7685388||Jan 25, 2008||Mar 23, 2010||Netapp, Inc.||Method and system for single pass volume scanning for multiple destination mirroring|
|US7702869||Dec 8, 2008||Apr 20, 2010||Netapp, Inc.||System and method for verifying the consistency of mirrored data sets|
|US7720801||Dec 17, 2004||May 18, 2010||Netapp, Inc.||System and method for supporting asynchronous data replication with very short update intervals|
|US7730071 *||Nov 28, 2006||Jun 1, 2010||Hitachi, Ltd.||Data management system and data management method|
|US7734951||Mar 20, 2006||Jun 8, 2010||Netapp, Inc.||System and method for data protection management in a logical namespace of a storage system environment|
|US7761424||Aug 10, 2006||Jul 20, 2010||International Business Machines Corporation||Recording notations per file of changed blocks coherent with a draining agent|
|US7769723||Apr 28, 2006||Aug 3, 2010||Netapp, Inc.||System and method for providing continuous data protection|
|US7882064||Jul 6, 2006||Feb 1, 2011||Emc Corporation||File system replication|
|US7925749||Apr 24, 2007||Apr 12, 2011||Netapp, Inc.||System and method for transparent data replication over migrating virtual servers|
|US7949843||Dec 7, 2009||May 24, 2011||Netapp, Inc.||Method and system for single pass volume scanning for multiple destination mirroring|
|US8001090||Nov 30, 2009||Aug 16, 2011||Netapp, Inc.||System and method for examining client generated content stored on a data container exported by a storage system|
|US8161007||Jan 27, 2010||Apr 17, 2012||Netapp, Inc.||System and method for supporting asynchronous data replication with very short update intervals|
|US8255504||Oct 3, 2006||Aug 28, 2012||United States Automobile Association (USAA)||Systems and methods for data source management|
|US8301791||May 30, 2008||Oct 30, 2012||Netapp, Inc.||System and method for non-disruptive check of a mirror|
|US8364645 *||May 7, 2010||Jan 29, 2013||Hitachi, Ltd.||Data management system and data management method|
|US8364920 *||Apr 6, 2009||Jan 29, 2013||Network Appliance, Inc.||System and method for transferring and backing up luns and lun clones on primary and secondary servers|
|US9015305||Aug 28, 2012||Apr 21, 2015||United Services Automobile Association (Usaa)||Systems and methods for data source management|
|US20040267823 *||Jun 24, 2003||Dec 30, 2004||Microsoft Corporation||Reconcilable and undoable file system|
|US20050027737 *||Jul 30, 2003||Feb 3, 2005||International Business Machines Corporation||Apparatus and method to provide information to multiple data storage devices|
|US20050144202 *||Dec 17, 2004||Jun 30, 2005||Chen Raymond C.||System and method for supporting asynchronous data replication with very short update intervals|
|US20060112151 *||Jan 9, 2006||May 25, 2006||Manley Stephen L||System and method for storage of snapshot metadata in a remote file|
|U.S. Classification||1/1, 714/E11.102, 707/E17.007, 714/E11.13, 707/999.205, 707/999.006, 707/999.101, 707/999.202|
|International Classification||G06F11/14, G06F11/20, G06F17/30|
|Cooperative Classification||G06F2201/84, G06F11/1471, G06F11/1451, Y10S707/99953, Y10S707/99942, Y10S707/99936, Y10S707/99956, G06F11/2082, G06F11/2064|
|European Classification||G06F11/14A12, G06F17/30C, G06F11/20S2S|
|Jan 5, 2009||FPAY||Fee payment|
Year of fee payment: 4
|Jan 7, 2013||FPAY||Fee payment|
Year of fee payment: 8